Blog - Investing Notes

October 3, 2006 - Testing the Statistical Significance and Economic Value of the Fed Model

Does the Fed Model work? In his late 2005 paper entitled "The FED Model and Expected Asset Returns", Paulo Maio examines the gap between the stock earnings yield and the 10-year Treasury note (T-note) yield to test both the predictive power and the economic value of the Fed Model. Using monthly data for stocks and bonds over the period 1954-2003, he concludes that:

• The yield gap correlates positively with future stock returns at both short (less than one year) and long forecasting horizons. Forecasting power for stock returns is greater at short horizons than at long horizons. At short (long) horizons, forecasting power is greater when the yield gap is negative (positive). Forecasting power for stock returns is greater in recessions than in expansions.
• The yield gap correlates negatively with future bond returns, at both short and long horizons. Forecasting power for bond returns is greater than when the yield gap is positive, and during expansions rather than recessions.
• Overall, the yield gap most reliably forecasts negative excess stock returns and positive excess bond returns.
• Changes in the yield gap correlate with with both discount rate news and cash flow news, confirming that the gap holds information about future earnings and returns.
• The yield gap has greater forecasting power for equally-weighted stocks than for value-weighted stocks, suggesting that it better predicts excess returns for small-capitalization stocks than for large-capitalization stocks.
• An out-of-sample investment strategy based on yield gap forecasting power generates higher Sharpe ratios than buy-and-hold strategies for both a broad stock market index and the T-note.

In summary, the Fed Model can help generate economically significant abnormal stock returns, most reliably within one year when predicting a bad market.

For related research, see Blog Synthesis: Gunning for the Fed Model? See also our Real Earnings Yield Model, which is Fed Model-like but compares the stock earnings yield with the inflation rate rather than the T-note yield.